This project will examine the role of social networks in the development and maintenance of heavy drinking by college students. A longitudinal research design will be used to probe two general processes: selection, in which students enter college as heavy drinkers and seek out other heavy drinkers to form a network; and social influence, in which the alcohol use of network members influences the drinking of the focal respondent. Special issues involving network composition will also be investigated (e.g., whether a romantic partner's or roommate's drinking exerts greater influence on the respondent than other members' alcohol use). Methodologically, the project will survey randomly selected freshperson dormitory residents from a large state university. The three assessment waves will be scheduled early in the fall semester of the student's first year, in the spring semester of their first year, and in the fall semester of their second year. A three-wave sample size of 200 complete cases will be sought. Measures will include: a standard social-network assessment grid; a variety of alcohol-related measures focusing on binge drinking (as defined in national studies by Henry Wechsler) but also including quantity-frequency, drunkenness, and alcohol problems; and measures of students' involvement in and attitudes toward aspects of college life (e.g., academics, parties). Data analyses will include structural equation modeling, path analysis, and latent growth modeling.